Persalys - Design of experiments

Persalys also provides the possibility of creating a design of experiments. Designs of experiments allows you to visualize the behaviour of your model thanks to scatter plots but also to build a metamodel that will then replace a costly physical model.

Several types of designs of experiments are provided:

Full factorial design
User-defined design
Probabilistic design
Imported design from a file

For the probabilistic design, Persalys offers three methods:

The LHS design (Latin Hypercube Sampling), optimised or not
Monte-Carlo design
Quasi-Monte-Carlo design

In addition to a summary of the statistic values of the sample, numerous displays of the data are available:

Probability density function or cumulative density function curve
Box plots
Parallel coordinates plots (cobweb plot)
Scatter plots 2d

It is possible to select the points in a view and to dynamically switch the view to see the selection from another tab.

For more information on how to create a design of experiments, refer to the following video: